Salesforce Agentforce: The Most Flexible Pricing Model in SaaS?
Plus: Commentary from Kyle Poyar, Scott Woody, Sam Lee, and Manny Medina.
Welcome back to Good Better Best!
Salesforce made some big updates to Agentforce last week that could have lasting implications on the future of SaaS pricing. I pulled in commentary from some of the sharpest pricing thinkers I know to help break down the changes, and what they mean.
Quick note before we get there: Next Friday, I’ll be jamming with Justin Farris, VP of Product at GitLab. We’ll be covering a range of topics, including:
Aligning Product and Pricing
Monetizing an open source product
Making Pricing reflect company values
But I want to hear from you. If you have specific questions, drop them in this thread, and I’ll make sure to add them to the agenda.
On to today’s post.
Your product is evolving—is your pricing keeping up?
Sam Lee, VP of Pricing Strategy & Product Operations at HubSpot shares hard-earned lessons on evolving from static pricing to usage-based models. This webinar covers hybrid pricing strategies, usage metric design, and how AI is accelerating the need for more elastic, value-aligned pricing. If you’re scaling product-led growth or navigating pricing transformation, start here.
Agentforce is all-in on Flexibility
Last week, Salesforce announced significant changes to Agentforce pricing. As one of the OG seat-based pricing companies, these shifts represent a noteworthy evolution in how AI services are being monetized in enterprise software.
I originally got wind of the news from
, CEO of Paid, a business engine for agents (and host of their awesome podcast, ), so I was especially curious to get his thoughts on the gravity of the announcement.When a $300B company like Salesforce abandons seat-based pricing for AI (even though they SWORE they wouldn't), it signals what we've been saying all along: the SaaS pricing playbook is dead for agents.
Strong words, and rightfully so!
So what exactly did Salesforce do? Let’s break it down below.
1. Flex Credits for Scalable Workflows
Salesforce introduced flex credits, offering customers a consumption-based model that better aligns costs with business outcomes. This approach allows businesses to pay only for the specific actions Agent Force performs, such as:
Updating customer records
Automating complex workflows
Resolving cases
Each action consumes 20 flex credits, with credits available in packs of 100,000 for $500. Customers maintain a Salesforce wallet to purchase these credits and optimize their AI spending.
Importantly, Salesforce already charged $2 per conversation for Agentforce (which is still an option).
This shift brings two significant improvements over the per-conversation model:
Greater predictability: Organizations can control exactly how many credits they purchase
Increased action visibility: Customers gain clarity on what specific actions they're paying for
This suggests Salesforce is pushing for more control and discipline, helping customers better understand their AI investments. But it also adds complexity. Here’s HubSpot’s Sam Lee:
The advantage of this model is twofold: First, it scales better with cost and value than conversation, especially for non-customer service user cases. Second, incentives are aligned between SFDC and its customers. Because cost scales with complexity and token volume, customers are incentivized to find the most efficient way to use Agentforce, leading to more intentional consumption.
The disadvantages? More complexity (and a lot of fine print). The model isn't bad when viewed in isolation, but SFDC's licensing is some of the most complex in the market. Flex Credit must live alongside all the various seats and (at least) two other credit models. Just thinking about how these different credit models may overlap and the resulting “compound metering” makes my head spin.
If it makes Sam’s head spin, I fear for the rest of us! But there’s more — Salesforce also announced a new spin on agreements.
2. Flex Agreements for Resource Allocation
Perhaps the most interesting innovation is the new flex agreement system, which allows organizations to manage both human and digital labor by shifting investments between user licenses and digital labor according to their priorities.
In practical terms, this means companies can convert user licenses into flex credits or vice versa. This creates a fluid boundary between paying for human employees versus AI agents.
Scott Woody, CEO of Metronome put this shift into perspective:
We’re seeing a clear trend: hybridization between seats and usage. Companies want predictability, but they also want pricing to reflect the actual value their customers get. For many, that means gradually evolving from pure seats into flexible, usage-aligned models. The belief is clear—long term, consumption and outcomes-based pricing is where things are headed. But the path to get there isn’t a clean switch. It’s hybrid, it’s messy, and it’s iterative.
From where we sit at Metronome, we’re seeing something new in the market: strong opinions, weakly held. Companies are launching with bold new pricing models, but they’re prepared to pivot fast if customer feedback or competitive pressure changes. This level of pricing and packaging agility hasn’t existed in decades. And it’s notable that Salesforce is leading this shift publicly. This is the company that practically invented seats and subscription pricing. That they’re moving toward usage is a signal of just how real this shift is. Credit to Benioff and the leadership team for walking the walk.
The very existence of this flexibility signals that AI agents aren't yet a "no-brainer" for the broader market. Salesforce is providing a middle ground for customers to experiment with the human-AI balance before fully committing to our agent overlords.
This agreement structure likely required significant backend development in Salesforce's pricing and billing infrastructure, and I’m genuinely curious how they’re pulling it off.
3. New User Licenses and Add-Ons with Included Usage
The final piece involves new Agent Force user licenses and add-ons that bring Agent Force capabilities to every employee. These offer unlimited employee-facing agent usage with a per-user-per-month pricing model.
This approach allows companies to include Agent Force usage within a seat license or create add-ons with included Agent Force use, complementing the flex agreement model while preserving options within the existing per-seat framework.
It may seem like Salesforce is offering every pricing option under the sun, and it kinda seems like that’s the case.
Here’s
on why companies are struggling to find footing with AI pricing:Seemingly everyone is struggling to find the best way to monetize AI products. Unlike traditional SaaS products, they:
Have real underlying costs (meaning margins can go pear-shaped with flat-rate pricing)
Can tap into headcount/FTE budgets, which are >>> than tech budgets
Are (somewhat) unpredictable in terms of outputs — sometimes AI works great, sometimes it doesn’t work at all
With that in mind, I’d expect more innovation on this model in the coming months. Regardless, here are my biggest takeaways so far:
Flexibility is paramount: The repeated emphasis on flexible credits, licenses, and agreements shows Salesforce wants to let customers scale Agentforce usage according to their unique needs.
The AI agent market is still maturing: If customers were fully comfortable with AI agents, this level of flexibility wouldn't be necessary. The shift toward credits suggests some buyer hesitation about fully embracing outcome-based models.
Pricing infrastructure will be a competitive advantage: The legacy software stack wasn't built to support this degree of pricing, billing, and contractual flexibility. How Salesforce is accomplishing this technically is an intriguing question. Tomorrow's winners will likely prioritize investments in pricing technology that enables dynamic pricing, billing, and contracting.
As AI continues to transform enterprise software, we can expect more innovation in how these services are packaged and sold. Salesforce's moves suggest the market is still finding its equilibrium between human and digital labor, with flexibility being the bridge to that future.
🎯 Expert Opinion
AI is fundamentally changing SaaS pricing, and at a different pace for different industries. The most successful companies understand their customer's value chain deeply, but they're pragmatic about where they attach pricing.
Here’s how I’m thinking about the bigger picture.
The Input vs. Outcome Divide
Pricing inputs (per-user or per-CPU-second) feels safe and familiar. But customers actively want to minimize inputs.
Charge per user, and they'll deploy the smallest possible team.
Charge for compute, and they'll optimize aggressively to reduce usage.
You're betting against efficiency. Outcomes tell a different story.
Price per model-in-production or per-prediction-made, and you're aligned with what customers want more of. Banks don't want fewer models—they want armies of them generating insights. When users are free, they'll add team members liberally, accelerating adoption.
Here's the thing: while outcomes offer better alignment and growth potential, markets aren't always ready. Even if a metric has the highest perceived value, customers may not have any expectation to pay for it.
Finding Your Fair Value Metric
The sweet spot is the "fair value pricing metric" something customers actively want more of and perceive as reasonable to pay for. Sometimes that's pure outcomes. Sometimes it's a hybrid bridging familiar inputs with emerging outputs (like what Salesforce is doing with Agentforce).
The key is flexibility. Start where the market is comfortable, then migrate toward outcomes as customers see value and develop new payment expectations.
Companies winning in AI aren't just building great technology; they're architecting pricing strategies that evolve with the market while staying true to value creation.
How we can help when you’re ready
Got questions? Join the free PricingSaaS Community to get fast answers from pricing leaders.
Leveling up? Check out our Resource Center for links to pricing and packaging research, webinars, and our free online course.
Need help? Our SaaS Pricing Assessment delivers a clear roadmap to smarter pricing and packaging in 30 days.